Polymerase θ-helicase efficiently unwinds DNA and RNA-DNA hybrids

POLQ is a unique multifunctional replication and repair gene that encodes for a N-terminal superfamily 2 helicase and a C-terminal A-family polymerase. Although the function of the polymerase domain has been investigated, little is understood regarding the helicase domain. Multiple studies have reported that polymerase θ-helicase (Polθ-helicase) is unable to unwind DNA. However, it exhibits ATPase activity that is stimulated by single-stranded DNA, which presents a biochemical conundrum. In contrast to previous reports, we demonstrate that Polθ-helicase (residues 1–894) efficiently unwinds DNA with 3′–5′ polarity, including DNA with 3′ or 5′ overhangs, blunt-ended DNA, and replication forks. Polθ-helicase also efficiently unwinds RNA-DNA hybrids and exhibits a preference for unwinding the lagging strand at replication forks, similar to related HELQ helicase. Finally, we find that Polθ-helicase can facilitate strand displacement synthesis by Polθ-polymerase, suggesting a plausible function for the helicase domain. Taken together, these findings indicate nucleic acid unwinding as a relevant activity for Polθ in replication repair.

POLQ is a unique gene in higher eukaryotes that encodes for a N-terminal superfamily 2 (SF2) helicase and a C-terminal A-family polymerase with a large central domain that lacks any known enzymatic domain ( Fig. 1A) (1)(2)(3). Understanding the biochemical activities and cellular functions of Pol has become a priority because it has been found to be essential for the error-prone double-strand break (DSB) 2 repair pathway known as microhomology-mediated end-joining (MMEJ) or alternative end-joining (alt-EJ) (3)(4)(5)(6)(7)(8)(9). Remarkably, Pol expression has also been shown to be important for the proliferation of cells deficient in the homologous recombination (HR) pathway, such as because of mutations in BRCA1 or BRCA2 (4,10). Recent studies additionally demonstrate that Pol is responsible for random DNA integration into the genomes of mammalian cells, and for T-DNA integration into plant genomes (11)(12)(13). In addition to these functions, Pol was shown to be essential for DSB repair in zebrafish embryos and is involved in replication timing and potentially replication fork repair (10,14,15). Thus, the recent expansion of Pol studies has revealed multiple essential and important functions for this enigmatic protein in DNA repair and cancer proliferation.
Although multiple studies have begun to elucidate the functions of the polymerase domain (Pol-polymerase), very little is understood about the helicase domain (Pol-helicase) which is a SF2 helicase member (Fig. 1A). For example, a seminal report investigating Pol activities found that the helicase exhibits ATPase activity as predicted from its conserved helicase motifs (i.e. nucleotide binding, single-stranded DNA (ssDNA) binding, and core helicase motifs) (Fig. 1A) (16). However, although Pol-helicase exhibits robust ATPase activity, the study failed to identify any DNA unwinding activity by the enzyme (16). Consistent with this, a more recent study reported that Pol-helicase is unable to unwind DNA (17). Interestingly, Ceccaldi et al. (10) reported that Pol-helicase interacts with RAD51 via specific binding motifs and exhibits anti-recombinase activity because of its ability to counter RAD51 activity. Despite these initial findings, the biochemical and cellular functions of Pol-helicase have yet to be fully elucidated.
Because Pol-helicase is most closely related to HELQ/ Hel308-type and RecQ-type helicases, it likely shares activities with these widely studied groups of motor proteins (1,18). For example, many RecQ helicases exhibit both DNA unwinding and annealing activities (19). Because these mechanisms can compete with one another, they can also mask each other in biochemical assays. For example, in recent studies we found that Pol-helicase exhibits DNA annealing activity, similar to RecQ-type helicases (20). Specifically, Pol-helicase promotes ssDNA annealing in an ATP-independent manner in the absence of the ssDNAbinding protein RPA (20). However, when RPA is prebound to ssDNA, Pol-helicase requires ATP hydrolysis to promote ssDNA annealing (20). These studies link the ATP-dependent annealing activity of the helicase to alt-EJ by showing that it counteracts RPA to promote end-joining (20) (reviewed in Ref. 21).
Because Pol-helicase promotes DNA annealing, we envisaged that this activity likely opposes its unwinding function, and if so this would explain why DNA unwinding by the helicase has been difficult to detect. Indeed, here we demonstrate that by masking ssDNA annealing, we observe that Pol-helicase efficiently unwinds several different types of DNA substrates with 3Ј-5Ј polarity, including replication forks, blunt-ended DNA, and DNA with 3Ј or 5Ј overhangs. We further demonstrate that Pol-helicase efficiently unwinds RNA-DNA hybrids and preferentially displaces the lagging strand from model replication forks, similar to the related HELQ/Hel308 helicase. These findings suggest Polhelicase DNA unwinding contributes to the many activities of Pol in genome maintenance, and highlight a new activity for this enigmatic multifunctional enzyme.

Pol-helicase unwinds DNA in an ATP-and dATP-dependent manner
Considering that Pol-helicase exhibits annealing activity like related RecQ-type helicases (20), it can conceivably rewind DNA after unwinding it, which would prevent detection of its unwinding function. We therefore developed an assay that would mask the annealing activity immediately following DNA unwinding by the helicase. Pol-helicase (residues 1-894) was expressed and purified from Escherichia coli using a N-terminal tandem hexahistidine-SUMO tag which was subsequently cleaved (Fig. 1B) (20). The purified helicase was incubated with a radiolabeled DNA substrate containing a 3Ј ssDNA overhang, referred to as partial ssDNA (pssDNA), in standard buffer conditions in the presence of MgCl 2 (Fig. 1C). Next, the ATPase activity of the helicase was initiated by adding ATP along with excess ssDNA trap that is identical to the short strand within the pssDNA substrate. Here, if the helicase unwinds the DNA duplex, then the excess unlabeled ssDNA trap will preferentially anneal to the complementary long strand within the pss-DNA substrate. Consistent with this, we detected helicase-dependent unwinding in the presence of the ssDNA trap (Fig. 1C), and show that excess sequence-specific ssDNA trap is essential for detection of Pol-helicase unwinding (Fig. S1A). To our knowledge, these data are the first to document Pol-helicase unwinding.
Next, we utilized the optimized unwinding assay to further characterize the enzyme's unwinding activity on various substrates. Unexpectedly, we observed that the helicase is able to unwind substrates containing 3Ј and 5Ј overhangs with similar efficiency (compare Fig. 1, C and D). Although related SF2 enzymes such as HELQ, also known as Hel308, translocate along ssDNA with a 3Ј-5Ј polarity (18), our data presented so far fail to reveal a particular polarity exhibited by Pol-helicase. Nevertheless, we proceeded to determine which nucleotide cofactors support the enzyme's unwinding activity. The results show that the helicase exclusively utilizes nucleotides containing adenine, but more efficiently unwinds DNA in the presence of ATP compared with dATP (Fig. 1E). We further find that the Pol-helicase is unable to unwind DNA in the presence of the nonhydrolyzable ATP analogue AMP-PNP which demonstrates that the enzyme harnesses the energy of ATP hydrolysis to unwind DNA as expected (Fig. 1F). Lastly, we demonstrate that Pol-helicase possessing a mutation of a highly conserved lysine (K121M) within the Walker A motif known to be essential for ATP binding fails to unwind DNA as expected (Fig. 1G). Taken together, the data presented in Fig. 1 clearly show that Pol-helicase exhibits robust DNA unwinding activity that depends on hydrolysis of ATP or dATP.

Pol-helicase preferentially unwinds DNA with 3 overhangs
Although Pol-helicase demonstrated a similar ability to unwind DNA containing a 3Ј or 5Ј ssDNA overhang, it would be unprecedented for such an enzyme to actively translocate along ssDNA in both directions. Thus, an alternative interpretation of the data presented in Fig. 1, C and D is that Pol-helicase actively translocates along ssDNA with a single polarity, but is capable of initiating unwinding at blunt or 3Ј recessed ends. To further investigate the enzyme's ATP-dependent directional movement, we assayed unwinding on 3Ј and 5Ј overhang substrates that contain longer DNA duplexes to increase the energy barrier to unwinding ( Fig. 2A). For example, the substrates used in Fig. 1 include a duplex region 15 base pairs (bp) in length, whereas the substrates used in the current figure contain 23 bp of double strand DNA. Importantly, the 23-bp duplex sequence on the 3Ј and 5Ј overhang substrates is identical to prevent differences in melting temperature, and thus the amount of energy required for unwinding. The results demonstrate that Pol-helicase unwinds the 3Ј overhang substrate, but not the 5Ј overhang substrate, indicating a 3Ј-5Ј polarity, similar to related HELQ/Hel308 ( Fig. 2A) (18).
The rate of unwinding by the helicase was next examined on multiple substrates to potentially identify its preference for a particular substrate. We utilized identical conditions to assay the enzyme on pssDNA containing 3Ј or 5Ј overhangs, duplex DNA, and a replication fork (Fig. 2, B-E). Here again, we employed substrates with the same double strand DNA sequence and thus identical melting temperature. The results show that although the helicase unwinds each substrate under identical conditions, it exhibits the highest rate of unwinding on pssDNA harboring a 3Ј overhang, which is consistent with 3Ј-5Ј directional movement along ssDNA (Fig. 2F). We presume the enzyme unwinds the replication fork at a slower rate because of a second enzyme acting on the 5Ј overhang that can conceivably impede helicase translocation on the 3Ј overhang. Taken together, the results presented so far in Fig.  2 demonstrate that Pol-helicase preferentially unwinds DNA containing 3Ј overhangs, but is also capable of unwinding double strand DNA, DNA with 5Ј overhangs, and replication forks. We note that although the enzyme can unwind blunt-ended DNA substrates, it fails to do so on longer substrates even at relatively high concentrations (Fig. S1B). This suggests that multiple Pol-helicase molecules are unable to act cooperatively to unwind long substrates, as indicated for SF1-type helicase UvrD (22). Because many helicases function with and are stimulated by the ssDNA binding protein RPA, we assessed whether RPA promotes Pol-helicase unwinding activity in Fig. 2G. Here, we determined the efficiency of unwinding the 23-bp duplex substrate by relatively low amounts of either Pol-helicase, RPA, or both proteins combined. The results show that the addition of both proteins results in synergistic activity which is indicated by a significantly higher yield of unwound DNA (Fig. 2G). Future studies will be required to determine whether RPA stimulation of Pol-helicase occurs by a specific protein-protein interaction.
Several lines of evidence indicate the involvement of RNA-DNA structures in contributing to both genome instability and DNA repair. For example, R-loops have long been associated with replicative stress and genome instability, whereas more recent work indicates that RNA-DNA hybrids can also promote DNA repair by mechanisms that remain to be elucidated (23)(24)(25)(26)(27). Considering the importance of RNA-DNA structures in DNA repair and genome instability, we proceeded to examine whether Pol-helicase unwinds RNA-DNA duplexes with similar efficiency. Indeed, using identical substrate sequences, our results show that RNA-DNA substrates are also efficiently unwound by Pol-helicase (Fig. 3, A and B).
Here again, the enzyme more rapidly unwinds the substrate containing a 3Ј overhang (Fig. 3A). We note that the helicase shows substantially lower efficiency of unwinding a blunt-

Pol-helicase unwinds DNA
ended RNA-DNA duplex (Fig. 3C). This is consistent with inefficient unwinding of a blunt-ended DNA-DNA duplex (see Fig. 2D). Failure of Pol-helicase to unwind a RNA-DNA substrate containing a 3Ј RNA overhang indicates that this enzyme exclusively translocates along ssDNA (Fig. 3D). The helicase also fails to unwind a RNA-RNA substrate, which further demonstrates its inability to translocate along RNA (Fig. 3E).

Pol-helicase efficiently unwinds substrates modeled after stalled replication forks
A previous report demonstrates that mammalian Pol acts in response to replication stress and promotes replication fork progression or fork stability (10). For example, Pol was shown to form cellular foci in response to ultraviolet light and confer cellular resistance to hydroxyurea treatment (10). Furthermore, Pol was demonstrated to promote replication fork progression in the absence of exogenous DNA damaging agents, and cells deficient in Pol exhibit a prolonged S phase delay and a significant increase in stalled or collapsed forks following hydroxyurea treatment (10). Thus, although Pol has an essential role in alt-EJ, additional lines of evidence suggest it might exhibit separate functions in response to replicative stress, such as replication fork restart (10).
We further examined Pol-helicase activity on different types of replication forks to provide insight into its potential functions during replication. Time courses of Pol-helicase unwinding were performed on replication forks containing leading or lagging strands, leading and lagging strands, or a fork lacking leading and lagging strands (Fig. 4A). The results clearly show that the helicase preferentially unwinds the fork containing the lagging strand but lacking the leading strand (Fig. 4A,  Fork B). These data suggest a possible function for Pol-helicase in replication fork repair. For example, following arrest of the leading strand polymerase, such as because of an encounter with a lesion, the replicative helicase is known to continue to unwind the fork, resulting in a large leading strand gap. In contrast, the lagging strand polymerase can continue to act on its respective template, generating Okazaki fragments (29). Hence, fork collapse is often modeled as Fork B which specifically lacks a leading strand. We next investigated whether Pol-helicase more efficiently unwinds the lagging strand which is common among Hel308-type enzymes. Indeed, similar to HELQ/Hel308 activity, we find that Pol-helicase preferentially unwinds the lagging strand at a replication fork which further supports a

Pol-helicase promotes strand displacement synthesis by Pol-polymerase
A unique feature of POLQ is that it is the only known gene in multicellular organisms to encode for a helicase and a polymerase. Other known helicase-polymerase fusion proteins are more common in bacteria, archaea, and viruses, and are involved in replication and repair (30). A conceivable function for Pol-helicase unwinding activity is to facilitate strand displacement synthesis by the Pol-polymerase domain. For example, although some polymerases exhibit proficient strand displacement activity, which enables DNA unwinding downstream of the 3Ј primer terminus during replication, many polymerases such as those involved in chromosomal replication require the unwinding activity of auxiliary helicases to perform replication of double strand DNA. We tested whether Pol-polymerase exhibits strand displacement activity on a replication fork containing a leading strand in Fig. 4C. The results show that the polymerase possesses limited strand displacement activity in the presence of all four dNTPs and ATP as indicated by its inability to fully extend the leading strand primer (Fig. 4C,  lane 2). Given that Pol-helicase exhibits 3Ј-5Ј polarity, we evaluated whether it promotes strand displacement activity by the polymerase domain. Indeed, addition of the helicase under identical conditions with ATP facilitates Pol-polymerase primer extension through the downstream DNA duplex, as indicated by a 4-fold increase in run-off product (Fig. 4C, lane  3). Hence, these data suggest a plausible function for the helicase domain in facilitating Pol-polymerase strand displacement synthesis during replication repair.

Discussion
Pol has multiple documented activities in DNA replication and repair, including alt-EJ, replication repair, translesion synthesis, and replication initiation (4 -8, 10, 15, 31, 32). Although the activities and cellular functions of Pol-polymerase have been investigated, little is understood regarding the enzymatic activities of Pol-helicase. For example, although studies have shown that the helicase exhibits ATPase activity that is stimulated by ssDNA, multiple reports failed to detect a DNA unwinding function which is common among DNA helicases sharing sequence homology to Pol, such as HELQ/Hel308and RecQ-type helicases (10, 16 -18, 28, 33-35). In this report, we demonstrate that Pol-helicase exhibits robust DNA unwinding activity with 3Ј-5Ј directionality. The helicase preferentially unwinds DNA substrates containing 3Ј ssDNA overhangs, but additionally unwinds substrates with 5Ј overhangs, blunt-ended DNA, RNA-DNA hybrids, and replication forks. Because Pol-helicase also performs ssDNA annealing (20), this function counters its unwinding which likely explains why Pol unwinding has been difficult to detect in previous studies (16,17).
Several lines of evidence have supported a role for Pol in replication fork repair. For example, a recent report demonstrated that suppression of Pol significantly slows the velocity of replication forks even in the absence of exogenous DNA damaging agents (10). This report also shows that knockdown of Pol expression impairs fork progression and halts cells in S-phase following hydroxyurea treatment (10). These data indicate that Pol either promotes replication fork stability or replication fork repair. Considering that several SF2 helicases, such as HELQ/Hel308 and RecQ subclasses, are involved in replication fork repair, it is not unreasonable to assume a similar function for Pol-helicase, which is closest in relation to HELQ/ Hel308 (18, 28, 36 -38). For example, prior studies showed that mammalian HELQ/Hel308 is recruited to stalled replication forks and is involved in repairing interstrand cross-links which arrest replication forks (28,37,38). Similar to HELQ/Hel308, Pol-helicase unwinds DNA with 3Ј-5Ј polarity and exhibits a preference for unwinding substrates modeled after collapsed replication forks, such as those lacking a leading strand (28,35). We also find that Pol-helicase is unable to unwind long substrates and thus exhibits nonprocessive unwinding activity like HELQ/Hel308 (18,35). Hence, our studies confirm similar biochemical activities between Pol-helicase and HELQ/Hel308 which suggests these enzymes perform similar replication repair functions.
Structural and sequence comparisons between Pol-helicase and Hel308-type enzymes provide further evidence for shared mechanisms of helicase activity (Fig. 5). For example, superposition of Pol-helicase and the co-crystal structure of Hel308 in complex with partially unwound DNA reveals a similar orientation of the ␤-hairpin motif, previously shown to act as a wedge to facilitate duplex unwinding by Hel308 (Fig. 5B) (39). Although the sequence of this motif is not closely conserved between Pol-helicaseand HELQ/Hel308-type enzymes (Fig.  5A), superposition of Pol-helicase and Hel308 suggests the slightly smaller ␤-hairpin in Pol facilitates DNA duplex separation by a similar mechanism (Fig. 5B). Another interesting structural similarity between these enzymes is the previously reported auto-inhibitory helix-loop-helix domain 5 which contains a highly conserved Arg-Ala-Arg (RAR) motif (Fig. 5C) (35,39). For instance, a prior report demonstrated that domain 5 within Hel308 suppresses its unwinding activity (35). Specifically, deletion of this domain or mutation of a conserved arginine (Arg-662) in this region, which was shown to interact with extruded ssDNA in the co-crystal structure of Hel308 in complex with partially unwound DNA, resulted in a dramatic increase in helicase activity (35,39). These types of helicase autoinhibitory domains found in both SF1 and SF2 members may be modulated by interacting proteins or specific nucleic acid structures (35). Thus, Pol-helicase activity may be substantially stimulated by protein or DNA interactions that change the orientation of the autoinhibitory domain. We speculate that the structurally and sequence conserved domain 5 in Pol-helicase exhibits an autoinhibitory mechanism like Hel308 (Fig. 5C).
Despite the similar unwinding activities between Pol and HELQ/Hel308, DNA unwinding is countered by the annealing function of Pol-helicase. For instance, detection of DNA unwinding by Pol-helicase requires masking the opposite annealing activity by addition of excess ssDNA trap. In contrast, HELQ/Hel308 has been shown to unwind DNA in the absence of a ssDNA trap and therefore does not likely exhibit strong annealing activity like Pol-helicase (28). Interestingly, other SF2 helicases, such as those from the RecQ subclass, Pol-helicase unwinds DNA exhibit ssDNA annealing, however, in many cases this activity is suppressed by ATP (19). In contrast, the respective annealing activities of Pol and RECQL5 helicases are not suppressed by ATP, and these enzymes share ϳ18% sequence homology (1,20). RECQL5 also unwinds DNA with low processivity like Pol-helicase (40). Despite these similarities, we were unable to detect strand exchange activity by Pol-helicase which RECQL5 has been shown to exhibit (Fig. S1C). Another similar function between Pol-helicase and RECQL5 is their ability to interact with RAD51 and counteract its activity. For example, both Pol-helicase and RECQL5 promote dissociation of RAD51-mediated D-loops in vitro, and these enzymes suppresses homologous recombination in cells (10,41). Taken together, Pol-helicase shares similar characteristics with RECQL5 and HELQ/Hel308. Because Pol is known to promote the proliferation of BRCA-deficient cancer cells and is considered a promising oncology drug target, it will be important to determine whether the unwinding function of the helicase domain contributes to cancer cell survival (4,10). For example, although the helicase domain was recently shown to promote alt-EJ via annealing and counteracting RPA, its unwinding activity may also contribute to this pathway (20). For example, Pol-helicase unwinding may enable microhomology annealing or strand displacement synthesis by Pol-polymerase during alt-EJ (Fig. 6A). Because Pol-polymerase exhibits poor strand displacement synthesis, DNA unwinding ahead of the polymerase would be a plausible function for the helicase during alt-EJ (Fig. 6A). Considering that RNA-DNA hybrids have recently been shown to form at DNA breaks, Pol-helicase dissociation of these structures may also contribute to DSB repair (27).
Importantly, it remains unclear whether alt-EJ independent roles for Pol-helicase exist and enable the proliferation of BRCA-deficient cells. For example, although a previous report suggested Pol-helicase suppresses homologous recombination via its RAD51 interaction motif (10), a more recent study was unable to confirm this mechanism in cells (20). Because Pol-helicase exhibits robust unwinding of replication forks, it can conceivably play a compensatory role in BRCA-deficient cells during replication repair. For instance, lagging strand unwinding can contribute to fork reversal and replication restart (Fig. 6B, left). Alternatively, the helicase can potentially promote replication by facilitating strand displacement synthesis by the polymerase domain (Fig. 6B, right). This activity could conceivably aid in replication restart by extending nascent primers. Future studies will be required to further characterize the molecular basis of Pol-helicase unwinding and determine whether this activity contributes to the many functions of Pol in replication and repair.

Pol-helicase unwinds DNA
KCl, 5% glycerol, 1 mM MgCl 2 ) then mixed with the indicated amounts of Pol-helicase for 5 min. This was followed by the addition of 2 mM ATP and 200 nM ssDNA trap for the indicated times at 30°C in a total volume of 20 l. Reactions were terminated by the addition of 4 l of non-denaturing stop buffer (0.2 M Tris-HCl, pH 7.5, 10 mg/ml proteinase K, 100 mM EDTA, and 0.5% SDS) then resolved in non-denaturing 12% polyacrylamide gels and visualized by phosphorimaging (Fujifilm FLA 7000) or autoradiography. For RPA stimulation experiments, the indicated amounts of RPA were pre-incubated with DNA for 5 min, then the indicated amounts of Pol-helicase were added for an additional 5 min. Reactions were then initiated as above. Unwinding experiments utilizing substrates with a 23-bp duplex were incubated at 37°C.

Sequence alignment
The indicated amino acid sequences of the helicase domain of Homo sapiens Pol and other indicated SF2/Ski2 family helicases were aligned using Clustal Omega (European Bioinformatics Institute) (http://www.ebi.ac.uk/Tools/msa/clustalo/) 3 default settings (44). Location and numbers of ␤-sheets and ␣-helices are indicated for Pol-helicase domain based on previous structural analysis (17).

Superposition of Pol-helicase and Hel308 structures
The C␣-bound form of Hel308 (PDB ID: 2P6R) was used as reference onto which the Pol-helicase domain (PDB ID: 5AGA) was superimposed using Swiss-PdbViewer (42). Using least squares fitting option, 1432 matching atoms were found to superimpose with a root mean square deviation (RMSD) of 1.55 Å. Images were generated with PyMOL software (43).

Pol strand displacement synthesis
10 nM 32 P-5Ј-radiolabeled DNA pre-incubated at room temperature in 25 mM Tris-HCl, pH 8.8, 1 mM DTT, 0.01% Nonidet P-40, 0.1 mg/ml BSA, 10% glycerol, 10 mM MgCl 2 was mixed with or without 50 nM Pol-helicase. Next, 2 mM ATP and 20 M dNTPs were added along with 400 nM unlabeled ssDNA trap for 30 min at 30°C. Pol-polymerase was added for an additional 20 min in a total volume of 20 l. Reactions were terminated by the addition of 20 l of 2ϫ denaturing stop buffer (90% formamide and 50 mM EDTA) then resolved in denaturing urea polyacrylamide gels and visualized by phosphorimaging (Fujifilm FLA 7000).